Inspire Brands is a multi-brand restaurant company whose current portfolio nearly 32,000 Arby’s, Baskin-Robbins, Buffalo Wild Wings (BWW), Dunkin’, Jimmy John’s, Rusty Taco, and SONIC Drive-In locations worldwide. The company was founded in 2018 and is headquartered in Atlanta, Georgia.
This project particularly focuses on the BWW's beer selection process. The brand sells beer in three ways: through bottles, cans, and taps, with the majority of beer sold from taps. When determining which beer brands to place on tap, the corporate locations’ product mixes across the nation are determined at the corporate level and consider a variety of factors, such as regionality, sales, and consumer preferences. Beer brands placed on tap range from nationally recognized brands to regionally recognized brands.
After the acquisition of Buffalo Wild Wings in 2018, the Inspire Brands management team identified a discrepancy in the brand’s proportion of total sales from beer compared to the industry standard for sports bars. The team stated that this difference stems from inconsistencies between beer supplied and demanded and a lack of data-driven insights within the beer selection process. The team expressed the need to move from a corporate-level selection process to a sports bar-level selection process to account for the growing shift in demand from national, well-recognized brands to regional, hyper-local brands. With these challenges in mind, two opportunities have been identified to address the design problems. First, create product demand forecasting on at the sports bar level, and second, streamline the beer selection process. Both opportunities are completed as a proof of concept based upon the sports bar locations in Georgia. By doing so, the proposed system will generate the most profitable product mixes on a sports bar-by-sports bar and month-by-month basis. The objective of this project is to increase profit in a two-fold process by 1) increasing revenue through capturing market potential and 2) decreasing costs through minimizing unused product.
To address the challenges, the design strategy is divided into three distinct stages: the training, processing, and solving stage. The first stage is the training stage. This stage utilizes a product demand forecasting model trained on historical sales data to understand which beers worked well on tap and why. The next stage is the processing stage. This stage utilizes the product demand forecasting model’s predictions to calculate the marginal value (i.e. additional profit) of adding a beer style to a product mix. Lastly, the solving stage utilizes an integer program optimization model that evaluates the marginal values for each beer style and solves for the product mix of highest profit. Lastly, the final output/deliverable is the monthly selection of beer styles at each sports bar, constrained by the number of taps available at a particular bar. The deliverable also contains the projected volume demanded for a beer during a given month. Each stage possesses its own validation approach, ranging from performance plotting to statistical analysis to sensitivity analysis.
The deliverable provided to the client encapsulates the three stages of the design strategy into a user-friendly application, called the Product Mix Selection Application. Once implemented, this application can utilize any inputted data from a given time period and region in the country. The Product Mix Selection Application, a locally-run desktop application, enables the Buffalo Wild Wings team to generate sports bar-specific monthly product mixes. The application’s interface utilizes human-computer interaction design principles and is minimalistic and intuitive. In addition, an accessible Help window, regular feedback on calculation progress, and carefully designed affordances aid in user adoption. In a series of button clicks, users have the ability to upload point of sales data, cost data, and promotional data to train the model. The last step outputs a CSV file of sports bar-specific monthly product mixes for the desired time period, including a range of values for expected demand.
Value and Impact
The value of the project is measured by comparing the proposed system’s projected profit to historical profit in 2019. After finding the difference between those two values, the proposed system observed, an increase in annual profit per sports bar, thus leading to a $"confidential" increase in annual profit across all corporate locations in Georgia. In regards to increasing profit, the proposed system observed a $"X" average annual increase in sales per sports bar by capturing market potential and thus providing customers with the beer they want to drink. Additionally, the proposed system observed a $"Y" average annual increase in costs per sports bar. Costs will increase when more demand is introduced into the system due to the fact that the system experiences such an increase in sales. Lastly, the proposed system reduces the time that it takes to select which beers to place on tap, moving from a six-month process to a two-week process. As the process is automated and streamlined, the need for assistance in the beer selection process will be minimized.